A recent study reveals over half of employees input high-risk information into generative AI tools, prompting urgent calls for stricter AI security policies.